The Breast Cancer Prevention Trial of the National Surgical Adjuvant Breast and Bowel Project was designed to ascertain the effect of tamoxifen on the primary prevention of breast cancer in women with at least moderate risk (1.66%, minimum 5-year risk) as defined by the Gail model. The trial was stopped early when a clear risk reduction for invasive breast cancer was demonstrated.1 However, a survival benefit was not defined, and quality-of-life measurements for major adverse outcomes were not performed. This omission has left gynecologists in a quandary regarding which women truly should be considered for chemoprevention.
Four studies using decision-analysis techniques have attempted to address questions generated by the early termination of the Breast Cancer Prevention Trial. Three of these reports did not include quality-of-life adjustments for potentially important health states (eg, stroke, postphlebitic syndrome, metastatic breast cancer).2–4 The authors of the fourth did incorporate quality adjustments but systematically underestimated tamoxifen benefit by assigning a quality-of-life score of 0.79 to the act of taking the tamoxifen pill.5 This low score was derived from a scenario that incorporated tamoxifen reduction of breast cancer risk related to patients with BRCA 1 and 2 mutations and not on an assessment of the adverse effects of tamoxifen. Therefore, the score of 0.79 likely represents a patient’s fear of the high remaining cancer risk rather than perceived problems with the medication.6 Because of the uncertainty stemming from the Breast Cancer Prevention Trial and subsequent analyses, the U.S. Preventive Services Task Force recently published recommendations for breast cancer chemoprevention that were limited in scope and specificity.7 They recommended against routine chemoprevention in women, regardless of breast cancer risk, but suggested that a discussion could take place between women at high risk of breast cancer and at low risk of tamoxifen-related adverse effects. This low-risk group was described as women who were “younger and have no predisposition to thromboembolic events or do not have a uterus.” Given that the qualifying “high” risk for breast cancer and the range representing younger age are not precisely defined, it is difficult to identify which patients even deserve this discussion. Most of the women represented by Breast Cancer Prevention Trial are relatively young and generally healthy. They will not be seeking preventive care advice from an oncologist but rather from their primary care physician. Therefore, in this article, our goal is to provide these physicians a decision analytic tool designed to
- More precisely define quality-of-life adjustments that reflect women’s perceptions of potential benefits and risks of tamoxifen chemoprevention.
- Incorporate the profound impact of stroke.
- Fully analyze the risk and quality-of-life effects of endometrial cancer in women who have not undergone hysterectomy.
- Propose some practical guidelines for physicians and patients who are considering chemoprevention.
MATERIALS AND METHODS
We interviewed a stratified sample of 106 women, aged 50 years or older, from urban areas of central North Carolina and south Florida. The stratification process was based on race and socioeconomic status. To achieve this goal, we recruited consecutive individuals at libraries, churches, medical clinics, and health fairs. These sites were chosen so we could obtain a sample that had a minimum of one-third African-American representation and a wide distribution of income and educational levels in both racial groups.
Questionnaires were administered verbally by a trained research assistant. The survey consisted of written scenarios designed to measure patient preferences for relevant health states. We used the standard gamble methodology, a valid measurement tool in health-economic analysis.8 Results were converted to health utility scores, for which 0 represented death and 1 represented excellent health. Using an alpha of .05 and 80% power to detect a 10% utility score difference between a 2-to-1 sample of whites to African Americans, we calculated a sample size requirement of 104 women assuming a standard deviation of 0.9. The health states studied were breast cancer with cure; endometrial cancer with cure; metastatic cancer with death in 2 years; stroke with mild, moderate, and severe debility; pulmonary embolism; deep venous thrombosis (DVT); and hot flushes. We obtained the utility scores for postphlebitic syndrome and hip fractures from previously published reports.9 Utility scores according to race were compared using t tests and linear regression analysis. Because we did not record subjects’ identities and because the research posed no medical risk to participants, this study was approved under the Moses Cone Health System institutional review board’s expedited review procedure.
We constructed a Markov State Transition Model10 designed to analyze tamoxifen chemoprevention in a hypothetical population of women representing the average patient in the Breast Cancer Prevention Trial.1 Markov models use published estimates of disease outcomes by transitioning patients to each outcome as appropriate in a set period of time, usually 6 months or a year. The time cycle repeats itself through the life expectancy of the cohort of patients of interest. In the absence of an appropriate randomized, controlled study, this process is able to simulate the cost and effect of a given treatment over time by incorporating known probabilities of each outcome and the cost of the treatment and its probable effects, whether good or bad.10 For the Breast Cancer Prevention Trial cohort, the average patient was a 50-year-old woman with an estimated 5-year breast cancer risk of 3.4%. Tamoxifen reduced this 5-year risk to 1.7%. As the cohort passes through a cycle, each individual accumulates a combination of time and quality attributable to her outcome for that cycle. This time-quality unit is expressed in quality-adjusted life-years. Note that postphlebitic syndrome, cerebrovascular accident with moderate debility, and cerebrovascular accident with severe debility are outcomes that involve permanent debility. Therefore, each subsequent year of survival is accrued with a quality decrement (eg, stroke with severe debility is rated with a health utility score of 0.20; therefore, each year of survival is only accumulated at the rate of 0.20 quality-adjusted life-years). The cycle length for each model transition was 1 year. The maximum number of cycles was set to average remaining life expectancy as defined by Centers for Disease Control and Prevention life tables.11 The transition outcomes that could occur in any given cycle were as follows: 1) well on tamoxifen therapy, 2) well off tamoxifen, 3) breast cancer, 4) DVT with and without postphlebitic syndrome, 5) pulmonary embolism, 6) endometrial cancer, 7) stroke, 8) breast cancer cure, and 9) death. Patients who suffered any complication of tamoxifen therapy were switched off of medication and were exposed to the probabilities of the “no tamoxifen group.” Patients who transitioned to breast cancer were placed on tamoxifen therapy and were exposed to the associated probabilities. Because the mean utility score for hot flushes approached 1 and because 85% of respondents reported no quality-of-life decrement, this outcome was not included. We did not incorporate bone fracture data in the base model because the Breast Cancer Prevention Trial data were not statistically significant and because another report suggested that, at the dose of 20 mg per day, tamoxifen did not alter fracture rates.12 However, we did include the trend toward reduced hip fracture incidence observed in the Breast Cancer Prevention Trial in the sensitivity analysis.
Subtrees were added for transition states as appropriate. The breast cancer branch was subdivided into tumor size categories as reported in the Breast Cancer Prevention Trial so that cure rates could be adjusted by tumor stage. The stroke branch was subdivided into degree of debility (mild, moderate, and severe) to allow for specific quality-of-life adjustments. All cancer branches were divided into 2 limbs, one representing cure and the other representing metastatic disease and then death.
Duration of reductions in health utility scores varied according to specific health states. For instance, the decrement for DVT absent postphlebitic syndrome and that of pulmonary embolism lasted 1 cycle because symptoms and complications for these states are temporary. However, the loss of quality of life associated with stroke debility and postphlebitic syndrome is permanent; therefore, utility decrements for these conditions lasted from disease onset until death. Previous studies have reported that women who experience breast cancer demonstrate prolonged reductions in quality of life even in those who are cured.13,14 Therefore, utility decrements for this group also lasted from disease onset until death.
For both the tamoxifen and placebo groups, the distribution of breast tumor size came directly from the Breast Cancer Prevention Trial, as did the annual probabilities of invasive breast cancer and tamoxifen side effects.1 Data on breast cancer survival were based on a series of reports published during the last decade.15–19 The probabilities of death attributable to each health state; the probability of each level of stroke debility; and the cost of care as a result of breast cancer, endometrial cancer, stroke, DVT, and pulmonary embolism were all extracted from the most-recent published estimates. (All probability assumptions, cost assumptions, and literature citations are in Table 1.) We converted cost data to 2002 U.S. dollars. The cost of tamoxifen was based on the Drug Topics Red Book28 listing for the generic preparation (average wholesale prices provided by the publishers of the Thomson Physician Desk Reference29) plus an 8% mark up to simulate retail pricing.
Women in the Breast Cancer Prevention Trial were limited to 5 years of tamoxifen because of data concerning adjuvant breast cancer therapy.30,31 A meta-analysis of these trials showed that, for the tamoxifen group, the occurrence of contralateral breast tumors was reduced nearly 50% for at least 10 years.32 Given that contralateral breast cancers essentially represent second primary tumors, it is extremely likely that the tamoxifen chemoprevention effect should persist in the same manner. Therefore, our baseline assumptions were 5 years of treatment for 10 years of durable effect. All analyses were performed using standard decision analysis software (Data 3.5; TreeAge Software, Inc., Williamston, MA).
The first part of the sensitivity analysis strategy was to ensure that many women beyond the average Breast Cancer Prevention Trial participant could reasonably be advised based on cost-effectiveness results. Given that life expectancy affects any cost-effectiveness calculation, models were adjusted for age. Also, the Breast Cancer Prevention Trial data did not show tamoxifen-related risk for DVT, pulmonary embolism, stroke, or endometrial cancer for women younger than age 50; therefore, we accounted for these risk reductions in the appropriate models. Because endometrial cancer risk represents a major concern in chemoprevention decisions, we performed separate analyses for women without a uterus.
The remainder of the sensitivity analysis was performed in the classic sense—to identify variables for which small changes produce large outcome effects and to prove the model robust through a wide range of assumptions. Variables and ranges tested in this part of the analysis are indicated in Table 1.
In the Breast Cancer Prevention Trial, tamoxifen treatment led to a relative risk reduction of approximately 50% in all groups tested. However, absolute risk reduction is the main factor that defines effectiveness. In addition to the average 5-year risk in the Breast Cancer Prevention Trial of 3.4% at baseline was reduced to 1.7% with treatment; we also tested 6% to 3%, 4.5% to 2.25%, 2.5% to 1.25%, 2% to 1%, and 1.6% to 0.8%.
Given that hip fracture often affects a woman’s longevity and quality of life,33,34 we performed sensitivity analyses using point estimates of hip fracture rates derived from Breast Cancer Prevention Trial. These analyses did not affect the result and are not discussed further.
The true effect of tamoxifen on stroke risk also is a concern. The Breast Cancer Prevention Trial data showed no stroke risk attributable to tamoxifen in study participants younger than 50 years of age. For patients aged 50 years and older, tamoxifen treatment was associated with a 69% increase in stroke probabilities (.0013 to .0022) that narrowly missed the .05 level for statistical significance. A recent review of tamoxifen side effects demonstrated a consistent increase in stroke for all tamoxifen trials.35 Given the potentially immense impact of stroke on quality of life, we included stroke risk calculations for patients aged 50 years and older. Past reports suggest a doubling of stroke risk for each decade beyond the age of 55,36 but the Breast Cancer Prevention Trial data were only dichotomized to younger than and older than age 50. Therefore, for patients aged 50 years and older, we performed an analysis using the Breast Cancer Prevention Trial stroke data and a second analysis using age-adjusted data.
Thirty-five percent of the 106 respondents were African American and 65% were white. The average age was 60 years. Other demographic characteristics of participating women were as follows: 54% were married; 18% had less than a high school education whereas 35% had greater than a high school education; and 11% were uninsured, 59% had private insurance, 29% had Medicare, and 1% had Medicaid. Only 9% of respondents felt their health was below average.
A woman representing the base case (50 years of age with uterus intact) had a remnant of 25.97 quality-adjusted life-years without tamoxifen compared with 26.07 quality-adjusted life-years with chemoprevention. The cost-effectiveness ratio was $43,300 per quality-adjusted life-year.
In Figure 1, we show quality-adjusted life-years and cost-effectiveness ratios according to age and the absence or presence of the uterus. Age, as expected, had a significant effect on quality-adjusted survival. In women aged 50 to 65 years, although tamoxifen always increased quality-adjusted life-years compared with no treatment, the high risk of endometrial cancer led to substantially higher costs and attenuated benefits. Given the low absolute risk of endometrial cancer, the cost of prophylactic hysterectomy was prohibitive ($100,000 per quality-adjusted life-year in the base model).
The cost of tamoxifen influenced all models, but because it is already available in generic form, we don’t expect any significant downward pricing in the U.S. market. Reducing the cost of breast cancer to $20,000 per lifetime compared with our assumption of $40,000 had little effect. However, tripling this estimate to $120,000 led to a 50% improvement in cost per quality-adjusted life-years gained in all tamoxifen cohorts. The average cost of stroke, endometrial cancer, and thromboembolic disease did not affect any model through the ranges tested.
We found no significant racial difference between white and African-American women for the utility scores of breast cancer. This result is important because the utility score for curable breast cancer influences cost-effectiveness calculations significantly. Refer to the Appendix for the standard gamble scenario used to measure this utility score. For example, in the base model, we found that if the utility score for curable breast cancer exceeded .9, the cost per quality-adjusted life-years eclipsed $50,000. For women older than 50 years, lower values assigned to life with curable breast cancer led to improved cost-effectiveness ratios. Importantly, if this utility score were less than .65 for a woman aged 60 years, the cost per quality-adjusted life-years for tamoxifen would drop beneath the $50,000 threshold. Variation in utility scores for other health states had no effect.
The probability of cure for early-stage breast cancer had no significant effect for the range tested. However, if we changed the cure rate for endometrial cancer from 80% to 60%, the cost of tamoxifen treatment rose to $93,300 per quality-adjusted life-years. The literature on tamoxifen does not support endometrial cancer cure rates that low.37
The baseline risk of breast cancer and, therefore, the absolute risk reduction attributable to chemoprevention were important in all models. Figure 2 demonstrates this effect. The starting risk of breast cancer per 5-year period is represented on the x-axis. The absolute risk reduction is equal to one half of this number. Gail model baseline probabilities coupled with patient age can be plotted on Figure 2 to illustrate the cost-effectiveness ratio for an individual patient. Note the horizontal line that demarcates the $50,000 per quality-adjusted life-year threshold.
Data from the Framingham Study clearly demonstrated that for women aged 55 years and older, the incidence of atherothrombotic stroke rises substantially with each decade.36 The annual incidence for women aged 55–64 years was reported to be 1.5 per 1,000. The incidence for women aged 65–74 years was reported to be 3.6 per 1,000. Given the intermediate position of age 60 in the reported range and the average annual incidence of stroke in the placebo group of the Women’s Health Initiative (2.1 per 1,000; mean age 63 years),38 we estimated annual stroke incidence in the 60-year-old “no tamoxifen” cohort to be 2 per 1,000. For the matching tamoxifen cohorts, we estimated a 69% increase in stroke risk as shown by the 50-year-old and older group in the Breast Cancer Prevention Trial. With the adjustments for higher stroke rates with age and standardizing 5-year breast cancer risk at 3.4%, the following results were obtained. For women aged 60 years with a uterus, the cost-effectiveness ratio was $128,000 per quality-adjusted life-years. For women aged 60 years without a uterus, this result was $63,000 per quality-adjusted life-years, which was still not cost-effective. The cost-effectiveness calculations represented by Figure 2 are adjusted for age-specific stroke rates.
Breast cancer chemoprevention has so far received a very limited endorsement from the U.S. Preventive Services Task Force7 and has been shrouded in pessimism by the press.39,40 This negative perception persists despite a recent report by Freedman et al41 that more than 10 million U.S. women qualify for breast cancer chemoprevention according to Breast Cancer Prevention Trial criteria and at least 2 million would experience a net benefit from treatment. A major reason for this skepticism is the difficulty in identifying women who would most benefit from this approach. By using the graph on Figure 2, this problem could be remedied. Adding prophylaxis to mammography screening produces 2 major advantages: 1) the incidence of breast cancer, the second-leading cause of cancer death among women,42 is actually reduced and 2) the quality-of-life decrements associated with disfiguring surgery, radiation, and chemotherapy are avoided.
The definition of an acceptable cost-effectiveness ratio has always been arbitrary, but $50,000 per quality-adjusted life-years gained as a result of a test or treatment has generally been regarded as a reasonable societal standard. In our analysis, tamoxifen chemoprevention, compared with no treatment, led to gains in remaining quality-adjusted life-years in all cohorts who had at least a 3.4% 5-year risk of breast cancer. However, cost-effectiveness only met the $50,000 cost per quality-adjusted life-year threshold for women aged 40 to 50 years. For women older than age 50, the increasing risk of stroke and endometrial cancer and the associated quality-of-life decrements make the cost of chemoprevention prohibitive unless a woman no longer has a uterus. For 60-year-old women who have experienced hysterectomy, the $50,000 cost-effectiveness ratio is achieved only if the 5-year baseline breast cancer risk is 4% or greater.
In most situations, it would be difficult to apply a decision model derived from a hypothetical cohort to individual treatment situations. However, this decision model remained robust through wide-ranging sensitivity analyses of multiple variables. The major exception was the health utility score for curable breast cancer. For women in their forties who have at least a 3.4%, 5-year risk of breast cancer, chemoprevention is cost-effective through the entire range of these scores. For a 50-year-old woman in the same risk group, if the utility score exceeds 0.9, then chemoprevention is not cost-effective. For women aged 60 years and older, chemoprevention could become cost-effective if this score is 0.7 or less. In other words, using our standard gamble scenario, a woman would have to be willing to take at least a 30% risk of dying immediately to avoid a breast cancer operation and associated therapies in the setting of guaranteed cure.
In terms of data that are available and the results reported here, the following chemoprevention generalizations can be made:
- For every woman who is aged between 40 and 49 years with at least a 3.4% 5-year risk of breast cancer, tamoxifen is cost-effective.
- For women aged 40 years, tamoxifen can be justified at a baseline 5-year risk as low as 1.5%.
- For women aged 50 to 65 years, the assessment of the utility score for curable breast cancer would be important because scores less than 0.7 would justify chemoprevention whereas those 0.9 or greater would always refute it.
- Assuming the utility score for curable breast cancer for women aged 60 to 65 years is average (0.8–0.85), tamoxifen would only be cost-effective if the baseline 5-year risk of breast cancer was very high (more than 5%).
- For patients who have a high probability of cerebrovascular disease (eg, known atherosclerosis, strong family history of stroke) or a hypercoagulable state, we would consider tamoxifen chemoprevention to be contraindicated.
Practically speaking, the 5-year breast cancer risk that would justify tamoxifen for women near the age of 40 years could be met by combining risk factors such as early age at menarche, late age of birth of first child, and 1 or more breast biopsies. For a woman aged 50 years or older, this baseline risk would not be reached unless she either had at least 1 first-degree relative with a history of breast cancer or had a personal history of a breast biopsy revealing atypical hyperplasia. Women with biopsy-proven ductal or lobular carcinoma in situ have a high enough risk of invasive breast cancer in all age groups to be considered.2 A role for tamoxifen in women who are BRCA-positive has yet to be defined. For computation of breast cancer risk, algorithms based on the Gail model are available in the original reports,2,43 and electronic versions can be obtained through the National Cancer Institute.7
In this study, we used Breast Cancer Prevention Trial data to estimate the effect size for adverse events and benefits attributable to tamoxifen. European prevention trials have not produced the risk reduction for invasive breast cancer reported in the Breast Cancer Prevention Trial,44–46 but all these trials allowed concomitant hormone replacement therapy (HRT) whereas the Breast Cancer Prevention Trial did not. Given that the Women’s Health Initiative demonstrated an odds ratio of 1.26 for breast cancer in the HRT group38 and that tamoxifen and HRT competitively bind to the same receptors,47 an attenuating effect on chemoprevention with cotherapy would not be surprising. The European trials were not powered to detect this effect. Reports on tumor prevention for the contralateral breast in women diagnosed with breast cancer, a group not allowed HRT, yielded similar RR reductions to the Breast Cancer Prevention Trial.32
In the recently reported International Breast Cancer Intervention Study I trial, the investigators reported a 2.8% 5-year rate of invasive breast cancer for the placebo group compared with 2.1% in the tamoxifen group, a relative risk reduction of only 25%.46 The most alarming finding in the International Breast Cancer Intervention Study I trial was the excess mortality in the tamoxifen group. The death rates for the placebo group (0.8 per 1,000) and the tamoxifen group (1.7 per 1,000) were well below population norms (3 to 4 per 1,000).11 The Breast Cancer Prevention Trial (2.7 per 1,000 for placebo and 2.2 per 1,000 for tamoxifen) and another report, the Italian Tamoxifen Study (7 per 1,000 placebo and 4 per 1,000 tamoxifen), reported opposite mortality trends.1,44 These data suggest that the International Breast Cancer Intervention Study I participants were unusually healthy and that rare mortal events were likely the result of chance. However, one caveat from International Breast Cancer Intervention Study I that should be explored is the association between tamoxifen use in the perioperative period and thrombotic complications. These data suggest either the perioperative discontinuation of tamoxifen or the institution of full DVT prophylaxis in this situation.
We have outlined specific circumstances in which tamoxifen for breast cancer prevention is cost-effective. Because of confounding with HRT and power issues in the European trials, we continue to regard Breast Cancer Prevention Trial as the most reliable source of efficacy data. Because of the stability of our sensitivity analyses and the robustness of our model, the only patient preference that needs to be ascertained to make clinical decisions is the health utility score for curable breast cancer in women aged 50 years and older. As a result, we hope that physicians and patients use this information as a guide. Tamoxifen is likely underused in selected patients younger the age of 50. For patients aged 60 years or older, chemoprevention should be rare. Because of the increasing risk of breast cancer in these older women, future work should focus on agents, such as selective estrogen receptor modulators and aromatase inhibitors, that affect the uterus and clotting system less. Until new data are available, tamoxifen remains the only choice for the primary prevention of breast cancer and should be restricted to women who are systematically identified.
1. Fisher B, Costantino JP, Wickerham DL, Redmond CK, Kavanah M, Cronin WM, et al. Tamoxifen for prevention of breast cancer: report of the National Surgical Adjuvant Breast and Bowel Project P-1 Study. J Natl Cancer Inst 1998;90:1371–88.
2. Gail MH, Costantino JP, Bryant J, Croyle R, Freedman L, Helzlsouer K, et al. Weighing the risks and benefits of tamoxifen treatment for preventing breast cancer. J Natl Cancer Inst 1999;91:1829–46.
3. Noe LL, Becker RV 3rd, Gradishar WJ, Gore M, Trotter JP. The cost effectiveness of tamoxifen in the prevention of breast cancer. Am J Manag Care 1999;5(suppl):S389–406.
4. Smith T, Hillner B. Tamoxifen should be cost-effective in reducing breast cancer risk in high-risk women. J Clin Oncol 2000;18:284–6.
5. Grann V, Sundararajan V, Jacobson J, Whang W, Heitjan D, Antman K, et al. Decision analysis of tamoxifen for the prevention of invasive breast cancer. Cancer J 2000;6:169–78.
6. Grann V, Jacobson J, Sundarajan V, Albert S, Troxel A, Neugot A. The quality of life associated with prophylactic treatments for women with BRCA 1/2 mutations. Cancer J Sci Am 1999;5:283–92.
7. US Preventive Services Task Force (USPSTF). Chemoprevention of breast cancer: recommendations and rationale. Ann Intern Med 2002;137:56–8.
8. Torrance G, Thomas W, Sacket D. A utility maximization model for the evaluation of health programs. Health Serv Res 1972;7:118–33.
9. Lenert L, Soetikno R. Automated computer interviews to elicit utilities: potential applications in the treatment of deep venous thrombosis. J Am Med Inform Assoc 1997;4:49–56.
10. Petitti D. Meta-analysis, decision analysis, and cost-effectiveness analysis: methods for quantitative synthesis in medicine. Monographs in epidemiology and biostatistics. Vol. 24. New York (NY): Oxford University Press; 1994. p. 142–47.
11. Anderson RN, DeTurk PB. United States life tables, 1999. National vital statistics reports; Vol. 50, no. 6. Hyattsville (MD): National Center for Health Statistics; 2002, p. 11–12.
12. Breuer B, Wallenstein S, Anderson R. Effect of tamoxifen on bone fractures in older nursing home residents. J Am Geriatr Soc 1998;46:968–72.
13. Ganz PA, Desmond KA, Leedham B, Rowland JH, Meyerowitz BE, Belin TR. Quality of life in long-term, disease-free survivors of breast cancer: a follow-up study [published erratum appears in J Natl Cancer Inst 2002;94:463]. J Natl Cancer Inst 2002;94:39–49.
14. Chie WC, Huang CS, Chen JH, Chang KJ. Utility assessment for different clinical phases of breast cancer in Taiwan. J Formos Med Assoc 2000;99:677–83.
15. Systemic treatment of early breast cancer by hormonal, cytotoxic, or immune therapy. 133 randomised trials involving 31,000 recurrences and 24,000 deaths among 75,000 women. Early Breast Cancer Trialists’ Collaborative Group. Lancet 1992;339:71–85.
16. Systemic treatment of early breast cancer by hormonal, cytotoxic, or immune therapy. 133 randomised trials involving 31,000 recurrences and 24,000 deaths among 75,000 women. Early Breast Cancer Trialists’ Collaborative Group [review]. Lancet 1992;339:1–15.
17. Fisher B, Dignam J, Tan-Chiu E, Anderson S, Fisher E, Wittliff J, et al. Prognosis and treatment of patients with breast tumors of one centimeter or less and negative axillary lymph nodes. J Natl Cancer Inst 2001;93:112–20.
18. Quiet C, Ferguson D, Weichselbaum R, Hellman S. Natural history of node-negative breast cancer: a study of 826 patients with long-term follow-up. J Clin Oncol 1995;13:1144–51.
19. Jacobson JA, Danforth DN, Cowan KH, d’Angelo T, Steinberg SM, Pierce L, et al. Ten-year results of a comparison of conservation with mastectomy in the treatment of stage I and II breast cancer. N Engl J Med 1995;332:907–11.
20. Will BP, Berthelot JM, Le Petit C, Tomiak EM, Verma S, Evans WK. Estimates of the lifetime costs of breast cancer treatment in Canada. Eur J Cancer 2000;36:724–35.
21. Taylor T, Davis P, Torner J, Holmes J, Meyer J, Jacobson M. Lifetime cost of stroke in the United States [review]. Stroke 1996;27:1459–66.
22. de Lissovoy G, Yusen R, Spiro T, Krupski W, Champion A, Sorensen S. Cost for inpatient care of venous thrombosis. Arch Intern Med 2000;160:3160–5.
23. Spirtos NM, Schlaerth JB, Gross GM, Spirtos TW, Schlaerth AC, Ballon SC. Cost and quality-of-life analyses of surgery for early endometrial cancer: laparotomy versus laparoscopy. Am J Obstet Gynecol 1996;174:1795–9.
24. Pourquier H, Gely S, Dubois JB, Joyeux H, Solassol C, Pujol H. Endometrial cancer: a comparative analysis of the therapeutic results and causes of failure after treatment by radiation combined with surgery or radiation therapy alone. Radiother Oncol 1990;17:115–22.
25. Brandjes DP, Buller HR, Heijboer H, Huisman MV, de Rijk M, Jagt H, et al. Randomised trial of effect of compression stockings in patients with symptomatic proximal-vein thrombosis. Lancet 1997;349:759–62.
26. Elliot MS, Immelman EJ, Jeffery P, Benatar SR. Funston MR, Smith JA, et al. A comparative randomized trial of heparin versus streptokinase in the treatment of acute proximal venous thrombosis: an interim report of a prospective trial. Br J Surg 1979;66:838–43.
27. van Beek E, Kuijer P, Buller H, Brandjes D, Bossuyt P, ten Cate J. The clinical course of patients with suspected pulmonary embolism. Arch Intern Med 1997;157:2593–8.
28. Drug topics red book. Montvale (NJ): Medical Economics;2002.
29. Physicians’ desk reference. 56th ed. Montvale (NJ): Medical Economics; 2002.
30. Fisher B, Dignam J, Bryant J, Wolmark N. Five versus more than five years of tamoxifen for lymph node-negative breast cancer: updated findings from the National Surgical Adjuvant Breast and Bowel Project B-14 randomized trial. J Natl Cancer Inst 2001;93:684–90.
31. Stewart HJ, Prescott RJ, Forrest AP. Scottish adjuvant tamoxifen trial: a randomized study updated to 15 years. J Natl Cancer Inst 2001;93:456–62.
32. Tamoxifen for early breast cancer: an overview of the randomised trials. Early Breast Cancer Trialists’ Collaborative Group. Lancet 1998;351:1451–67.
33. Cummings S, Melton L. Epidemiology and outcomes of osteoporotic fractures [review]. Lancet 2002;359:1761–67.
34. Tosteson AN, Gabriel SE, Grove MR, Moncur MM, Kneeland TS, Melton LJ 3rd. Impact of hip and vertebral fractures on quality-adjusted life years. Osteoporos Int 2001;12:1042–9.
35. Shapiro C, Recht A. Side effects of adjuvant treatment of breast cancer [review]. N Engl J Med 2001;344:1997–2008.
36. Wolf P, D’Agostino R. Epidemiology of stroke. In: Barnett H, Mohr J, Stein B, Yahtsu F, editors. Stroke: pathophysiology, diagnosis, and management. 3rd ed. Philadelphia (PA): Churchill Livingstone; 1998. p. 3–28.
37. Bergman L, Beelen ML, Gallee MP, Hollema H, Benraadt J, van Leeuwen FE. Risk and prognosis of endometrial cancer after tamoxifen for breast cancer. Comprehensive Cancer Centres’ ALERT Group. Assessment of Liver and Endometrial cancer Risk following Tamoxifen. Lancet 2000;356:881–7.
38. Rossouw JE, Anderson GL, Prentice RL, LaCroix AZ, Kooperberg C, Stefanick ML, et al. for the Writing group for the Women’s Health Initiative Investigators. Risks and benefits of estrogen plus progestin in healthy postmenopausal women. JAMA 2002;288:321–33.
39. Schwartz LM, Woloshin S. News media coverage of screening mammography for women in their 40s and tamoxifen for primary prevention of breast cancer. JAMA 2002;287:3136–42.
40. Grady D. Breast cancer drug is a uterus cancer risk. The New York Times. April 14, 1998;sect. Science:1.
41. Freedman A, Graubard B, Rao S, McCaskill-Stevens W, Ballard-Barbash R, Gail M. Estimates of the number of US women who could benefit from tamoxifen for breast cancer chemoprevention. J Natl Cancer Inst 2003;95:526–32.
42. Edwards BK, Howe HL, Ries LA, Thun MJ, Rosenberg HM, Yancik R, et al. Annual report to the nation on the status of cancer, 1973–1999, featuring implications of age and aging on U.S. Cancer Burden. Cancer 2002;94:2766–92.
43. Gail MH, Brinton LA, Byar DP, Corle DK, Green SB, Schairer C, et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst 1989;81:1879–86.
44. Veronesi U, Maisonneuve P, Sacchini V, Rotmensz N, Boyle P. Tamoxifen for breast cancer among hysterectomized women. Lancet 2002;359:1122–4.
45. Powles T, Eeles R, Ashley S, Easton D, Chang J, Dowsett M, et al. Interim analysis of the incidence of breast cancer in the Royal Marsden Hospital tamoxifen randomised chemoprevention trial. Lancet 1998;352:98–101.
46. Cuzick J, Forbes J, Edwards R, Baum M, Cawthorn S, Coates A, et al. for the IBIS investigators. First results from the International Breast Cancer Intervention Study (IBIS-I): a randomised prevention trial. Lancet 2002;360:817–24.
47. Lippman M, Monaco M, Bolan G. Effects of estrone, estradiol (E2), and estriol on hormone-responsive human breast cancer in long-term tissue culture. Cancer Res 1977;37:1901–7.
In this appendix, we have reproduced a standard gamble scenario that was used to elicit health utility scores for curable breast cancer. Scenarios describing lumpectomy plus radiation therapy and simple mastectomy were highly correlated and both yielded mean utility scores of 0.83. Therefore, for women aged 50 to 65 years who meet the 5-year breast cancer risk criteria for consideration of tamoxifen chemoprevention, we would simply recommend using the mastectomy scenario below. Note that the point of indifference (the patient discerns no difference between the guaranteed health state of mastectomy with cure and the immediate risk of death from the treatment that avoids this state) subtracted from 1 yields the utility score. For example, if a woman says that rather than undergoing mastectomy with guaranteed cure she would regard a 10% risk of dying right now as equivalent to the avoidance of mastectomy, her utility score for breast cancer with cure would be 0.9. If the point of indifference is 10% or less (yielding a utility score of ≥ 0.9), tamoxifen should not be considered.
You have recently been given a diagnosis of breast cancer. The cancerous lesion is small and located in only 1 breast. It has been determined that the cancer will be cured if the breast with the cancer is entirely removed. The only discomfort involved will be the local pain after surgery and then the dressing changes and healing process for 1 month after surgery. To avoid the surgical removal of 1 breast, there is a treatment available that will cure the cancer without having this surgery and postoperative pain. However, the treatment can cause immediate death. What risk of dying right now are you willing to take to cure the breast cancer, avoid the surgical removal of 1 breast, and prevent the 1 month of postoperative pain and dressing changes?